A frame-semantic model of Czech experiential expressions

Focusing on the ways in which Czech speakers talk about human experiences, I  show that experiential expressions go well beyond lexical predicates of perception, emotion, or cognition and that the issues of argument-coding involve much more that the often discussed 'psych-verb' phenomena or dative experiencers. I argue that the conceptual apparatus of Frame Semantics (Fillmore 1982, Fillmore & Atkins 1992, Atkins 1994) and Construction Grammar (Fillmore & Kay 1995) can offer particularly useful analytic tools for formulating adequate generalizations about the complex relationships between meaning and form within this traditionally difficult area. I analyze a rich collection of patterns, ranging from predicates of experience (controlled or uncontrolled) and certain causatives (prekvapit), to complex constructions involving causation (picha ho v boku; ovanulo ho chladem), reflexive constructions (spalo se mi dobre), and swarm-like alternations (jejich dum (mi) smrdel cesnekem). Many of these expressions involve predicates that are inherently non-experiential, thus posing the question of what allows the speakers to successfully generate and interpret these patterns as experiential.

My approach is based on the assumption that (i) argument expression is motivated by speakers' construal of events and (ii) the same piece of reality can be construed in multiple ways, thus giving rise to multiple instantiations of the same participant. I propose that the patterns at hand are organized in a coherent semantic network, unified by a shared background frame called 'Human Experience' (HE). This frame represents a generalized knowledge structure containing the participants 'experiencer', 'stimulus', and 'domain'. Different formal expressions then reflect different subsets and configurations of these three participants.

In existing research, frames have been used mostly as a tool for defining predicate classes or predicates whose lexical meaning can be structured and represented in terms of a shared background scene (e.g. buy/sell/cost/etc. all sharing a 'commercial event' frame). I take this approach a step further and argue that there is evidence for frames as semantic entities that speakers can use independently of specific lexical verbs. Thus, experiential construal can be either inherently associated with certain predicates, or it can be 'built up' gradually, by integrating inherently non-experiential predicates with the experiential background frame, thus allowing speakers to create semantically more complex construals involving HE. Crucially, the construals are not always compositional, nor can they be always treated simply as metaphorical extensions.

The main points of the present analysis can be summarized as follows. It

* provides a cognitively plausible unifying semantic structure that holds together formally diverse but semantically related expressions;

* highlights the need to maintain the distinction between lexical predicates and larger construal patterns (a number of otherwise inexplicable relationships can be attributed to the tension between the two layers of semantic structure);

* offers motivation for various constructional overlaps attested in the data (dative linking, agent-demotion, reflexivization);

* establishes semantic constraints on what predicates classes or grammatical constructions can unify with the semantic structure contributed by the HE frame (semantic compatibility is crucial);

* easily accommodates the fact that there is a fuzzy line between lexical verbs, complex lexical items, and truly productive phrasal expressions (the Czech material contains examples of all three categories).